Nvidia's moat moves beyond GPUs to orchestration
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Nvidia is rolling out its Vera Rubin architecture, pairing the Rubin GPU with the Vera CPU, Groq 3 LPX inference accelerator, and dedicated storage and networking racks designed to orchestrate data movement rather than churn tokens.
- Jason Hardy, Nvidia's VP of storage technology, said the Vera CPU delivers up to 3x improvement in storage operations by eliminating bottlenecks between flash memory and GPU processing — 'now we can use our flash to its fullest potential.'
- Nvidia shares grew roughly 10x between early 2023 and mid-2025, but the stock has since tracked modestly as Amazon and Google built competing in-house chips, prompting investor questions about GPU durability.
- OpenAI's Jalapeño chip targets the same problem via a different route — keeping whole workloads within one large integrated domain to minimize data movement and communication delays.
- The new competitive layer is systems-level efficiency rather than GPU performance, and Nvidia appears to hold a commanding lead in the early stages of this shift.
Why it matters: Investors who treated Nvidia as a pure GPU play facing hyperscaler competition are now confronting a company extending its lead into the orchestration and data-movement layer as compute scales to gigawatts. Hardy attributed up to 3x storage-operation gains to the Vera CPU, suggesting Nvidia's systems stack — not its GPU monopoly — is what defends pricing power against Amazon's and Google's in-house silicon.
Ask SkimNews
